Loose filament tangled above a printer’s bed gives a camera a conspicuous clue. A developing nozzle clog may offer very little to see. That difference helps explain the reach of Obico’s AI print monitor.[1]
Obico describes a computer-vision model trained on normal and failed prints. Its plugin takes webcam snapshots; the model assigns a failure-confidence score. Crossing a chosen threshold can trigger a notification and, if configured, a pause command.[1]
The company identifies spaghetti—the extrusion of loose, tangled strands—as its strongest detection category. It also watches for detachment, shifted layers and large nozzle blobs. A side view makes layer shifts easier to see; looking straight down can conceal them. Clogs and under-extrusion may remain hard to recognise until their effects become substantial.[1]
Sensitivity sets how readily the system acts on that visual evidence. Lowering it reduces false alerts but can miss real failures. Stable mounting and consistent, diffuse light improve the images reaching the detector before that threshold is applied.[1]
Sources & further reading
- How Obico AI Failure Detection Works Obico · Accessed 20 Oct 2026